The Prevalence of Titanium Dioxide Particles in Synovial Fluid Samples Drops after European Union Ban

: Due to health concerns, the European Union has banned the use of titanium dioxide nanoparticles in consumables in February 2022, with a 6-month transitional period ending in August 2022. We studied the prevalence of titanium dioxide nanoparticles in synovial fluid samples during and after the transitional period. A total of 302 samples were collected as a consecutive series between 1 April 2022 and 15 June 2023 from patients visiting the department of rheumatology at VieCuri Medical Centre in Venlo, The Netherlands. The samples were primarily collected for diagnostic purposes and only clinical waste material was used for this study. From each sample, up to 40 µ l of fluid was analysed with Raman spectroscopy for the presence of titanium dioxide particles. The trend in prevalence was calculated with a 3-month wide moving average. A total of 13 out of 302 samples (4.3%) contained titanium dioxide (TiO 2 ). The prevalence of TiO 2 decreased between the transitional period and the period after the ban ( p = 0.0154, with a relative risk ratio of 4.9 (95% CI 1.35–17.74). There was no significant difference in patient characteristics between the TiO 2 positive and the TiO 2 negative group. These results are hinting towards a possible relationship between the EU-ban and the identified decrease in prevalence.


Introduction
Titanium dioxide (TiO 2 ) is known worldwide as a ubiquitous element of foods, drugs, ointments, toothpaste, and other consumables [1].It is mostly used as a whitening agent, with particle sizes ranging from 20 to 400 nm, although most identified particles are <100 nm [2,3].Due to their small size, TiO 2 particles can cross biological barriers and are therefore already found in deep tissues, such as the temporal lobe of patients with Alzheimer's disease [4].Although previously considered safe, concerns on possible carcinogenic and pro-inflammatory properties have been reported [2,5].The adverse effects of TiO 2 are predominantly caused by oxidative stress reactions [5,6].The exact retention time of TiO 2 in the human body is unknown but estimates of tissue half-life range from 28 to 650 days for rats, dependent on the type of particle and organ [7].The health concerns eventually led to a ban on the use of TiO 2 for human consumption in food in the European Union [8].The ban came effective on 7 February 2022, with a 6-month transitional period ending August 2022.
Next to their presence in the brain stem [4], TiO 2 particles have also been identified in synovial fluid samples from patients with swollen joints, including in an ankle of a patient with a suspected gout flare [9,10].Synovial fluid is a protein-rich viscous fluid present as a thin layer in joint spaces.Particles smaller than 100 nm are known to be able to potentially pass the synovial membrane and can therefore infiltrate the joint space [11].This is the most likely pathway for TiO 2 to be present in native joints, although TiO 2 is also hypothesized to be shed in the synovial space due to wear-and-tear of orthopedic implants [12].Particle-associated inflammation is common in the human joint, due to the presence of pathological crystals which can induce arousal of the innate immune system.Gout, related to the presence of monosodium urate (MSU) crystals in the synovial fluid, is even the most common form of inflammatory arthritis in humans [13].Like MSU crystals [14], TiO 2 nanoparticles can mediate inflammation through the NLRP-3 (NOD-, LRR-and pyrin domain containing protein 3) inflammasome.Co-stimulation of THP-1 macrophages with LPS (lipopolysaccharides) and TiO 2 nanoparticles led to an increase in secretion of pro-inflammatory cytokines interleukin 1β, interleukin 6, and TNF-α (tumour necrosis factor) [15].This mechanism is an argument for a potential role of TiO 2 in the induction or progression of inflammatory joint disease.
In this study, we investigated the prevalence of TiO 2 particles in synovial fluid samples from patients visiting the outpatient rheumatology clinic in Venlo, The Netherlands, during the period shortly after the ban on these particles became effective.All patients visited for routine clinical care and were suffering from a swollen joint or joint bursa.We retrieved the samples by arthrocentesis, subsequently using polarized light microscopy to locate birefringent material (TiO 2 is birefringent [16]), followed by a characterization of particles with Raman spectroscopy.

Study Design and Participants
This study was designed as a single-centre, inception cohort study.A consecutive series of synovial fluid samples from patients who visited the outpatient clinic of the department of rheumatology from VieCuri Medical Centre (Venlo, The Netherlands) between 1 April 2022 and 15 June 2023 was collected.Eligible for inclusion were patients above the age of 18 with any swollen peripheral joint who underwent arthrocentesis as part of their routine clinical care.All samples have been reviewed by a rheumatologist (>20 years of experience) with polarized light microscopy for the presence of pathological crystals, such as monosodium urate for gout.
Only clinical waste material was used in this study and clinical data was retrieved from electronic health records.Patients gave written informed consent to the use of their synovial fluid and data.The study was performed according to IGH/GCP guidelines and the Helsinki declaration.The hospital board of directors from VieCuri Medical Centre reviewed and approved the study protocol.

Test Methods
Fluid was drawn from patients using a BD Plastipak syringe (Beckton Dickinson, Franklin Lakes, New Jersey, USA).From each patient sample, a droplet of up to 40 µL of synovial fluid (dependent on joint) was placed on a standard microscope slide (Epredia, MI, USA) and covered with a coverslip (Menzel-Gläser, Braunschweig, Germany).The sample was then placed under the Hybriscan iRPolM integrated Raman polarized light microscope (Hybriscan Technologies B.V., Nijkerk, The Netherlands), which operates a 20× 0.8 NA objective.Birefringent objects were located manually (TN), which were then scanned with the integrated Raman spectroscope.For each object, the Raman spectrum from 0 to 3600 cm −1 was measured for 1 s/pixel (Raman Spectrum) with a laser power of approximately 10 mW and a laser with an excitation wavelength of 532 nm.Dependent on crystal size, 60 to 120 pixels were measured, resulting in a total measurement time of 1-2 min.The spectra were compared using visual plot examination with earlier reported spectra of anatase and rutile [17].A patient was considered positive for TiO 2 if at least one birefringent object could be identified as such by Raman Spectroscopy.
Fifteen blank samples were performed on a random sample of the BD Plastipak syringes used for collection by rinsing with analytical grade MiliQ ® , which was stored overnight in the syringe.Per syringe, 40 µL of liquid was placed on a microscope slide and directly covered with a 18 × 18 mm cover glass.Blank samples were analysed with a similar method as used for the patient samples.
Before each sample, the Raman spectroscope is calibrated with (sequentially) CCD camera offset measurements, CCD camera dark current measurement, a Neon calibration standard, a PMMA calibration standard, a NIST intensity calibration measurement and a background (air) measurement.

Analysis
The differences between the group TiO 2 positive patients and the group TiO 2 negative patients were compared using chi-square tests (sex, analysed joint, presence of orthopaedic implants, clinical diagnosis, C-reactive protein levels) and two-tailed student T-test (age).The trend in prevalence over time was calculated with a 3-month wide moving average (MOVAG), including 95% confidence intervals, with Python 3.10.9.Using the SciPy toolkit, an exponential function was fit using least-squares regression.The significance in the difference of titanium dioxide prevalence between the transitional period from 7 February 2022 to 7 August 2022 and the period after the ban was tested with a chi-square test, and the Altman Relative Risk ratio.The alpha level for all tests was set to 0.05.

Results
In total, 302 synovial fluid samples were analysed.iRPolM analysis was on average performed within one day after aspiration (SD ± 1.7).A total of 13 patients (4.3%) were positive for TiO 2 particles in their synovial fluid.None of the blank samples were positive for TiO 2 according to the test.The count of particles per patient was low (ranging from 0 to 3 particles per 40 µl sample).The average size of the particles was 4.4 (SD ±1.4) µm.We identified two polymorphs of TiO 2, which were anatase and rutile.Seven patients had anatase particles in their fluid, and six patients were positive for rutile particles.Examples of the Raman spectra from both particle types can be found in Figure 1.We never identified both polymorphs in the same patient.
overnight in the syringe.Per syringe, 40 µL of liquid was placed on a microscope slide and directly covered with a 18 × 18 mm cover glass.Blank samples were analysed with a similar method as used for the patient samples.
Before each sample, the Raman spectroscope is calibrated with (sequentially) CCD camera offset measurements, CCD camera dark current measurement, a Neon calibration standard, a PMMA calibration standard, a NIST intensity calibration measurement and a background (air) measurement.

Analysis
The differences between the group TiO2 positive patients and the group TiO2 negative patients were compared using chi-square tests (sex, analysed joint, presence of orthopaedic implants, clinical diagnosis, C-reactive protein levels) and two-tailed student T-test (age).The trend in prevalence over time was calculated with a 3-month wide moving average (MOVAG), including 95% confidence intervals, with Python 3.10.9.Using the SciPy toolkit, an exponential function was fit using least-squares regression.The significance in the difference of titanium dioxide prevalence between the transitional period from 7 February 2022 to 7 August 2022 and the period after the ban was tested with a chi-square test, and the Altman Relative Risk ratio.The alpha level for all tests was set to 0.05.

Results
In total, 302 synovial fluid samples were analysed.iRPolM analysis was on average performed within one day after aspiration (SD ± 1.7).A total of 13 patients (4.3%) were positive for TiO2 particles in their synovial fluid.None of the blank samples were positive for TiO2 according to the test.The count of particles per patient was low (ranging from 0 to 3 particles per 40 µl sample).The average size of the particles was 4.4 (SD ±1.4) µm.We identified two polymorphs of TiO2, which were anatase and rutile.Seven patients had anatase particles in their fluid, and six patients were positive for rutile particles.Examples of the Raman spectra from both particle types can be found in Figure 1.We never identified both polymorphs in the same patient.There was no significant difference in clinical diagnosis, sex, age, analysed joint, and serum C-reactive protein levels between the TiO 2 positive patients and TiO 2 negative patients.Two patients with TiO 2 particles had a history of orthopaedic implants: one patient was punctured in a knee with a total knee prosthetic, and one patient was punctured in their wrist, whilst having an osteosynthesis plate in their adjacent radius.Full patient characteristics can be found in Table 1.The prevalence of TiO 2 in the synovial fluid samples was not constant but changed over the collection period, as visible in the timeline plot (Figure 2).A total of 114 samples were measured between 1 April 2022 and 7 August 2022, of which 9 were positive for TiO 2 .A total of 188 samples were measured between of 8 August 2022 and 15 June 2023, of which 3 were positive for TiO 2 .There is a clear decreasing trend visible, and from November 2022, no TiO 2 particles were identified anymore.The prevalence of TiO 2 in synovial fluid samples was higher in the transitional period than during the period with the ban in effect (p = 0.0154, with a relative risk ratio of 4.9, 95% CI 1.35-17.74).A total of 302 samples from patients with swollen joints were analyzed with a H-iRPolM Raman Spectrometer for the presence of TiO2.All patients were recruited from the VieCuri Medical Centre in Venlo, The Netherlands.On average, samples were analyzed within 24 h after puncture (SD 1.5 day).The trend in prevalence was calculated with a 3-month MOVAG algorithm, which was then used to fit an exponential function (  =  ) using least squares regression.There was a significant difference in the prevalence of TiO2 between the transitional period (February 2022-April 2022) and the period during the ban (p < 0.01).

Discussion
We here present a decrease in prevalence of TiO2 nanoparticles in synovial fluid samples drawn from Dutch patients with swollen joints.This decrease overlaps in time with the new EU regulations banning the use of TiO2 nanoparticles in food products.Although TiO2 was never identified in patients with active gout, simultaneous presence of TiO2 particles and calcium pyrophosphate crystals was identified in two patients.The question rises whether a synergistic effect can be attributed to CPP and TiO2, as both particles can induce arousal of the NLRP3 inflammasome [15,18].The combination of two pro-inflammatory particles in the same inflamed joint is known as mixed crystal disease [19], a condition mostly attributed to the simultaneous presence of MSU and CPP crystals.As this is a rare phenomenon, there is little known about the severity of the disease and optimal disease management.
This study is not without limitations.Raman spectroscopy is a qualitative method and relies on microscopic localization of particles.Optical microscopic platforms are limited in resolution to the diffraction limit and have difficulties distinguishing objects smaller than ~300 nm.Furthermore, we analyzed only a small volume (40 µL) of fluid, whilst synovial fluid punctures can retrieve up to 50 mL of material.The patients were all recruited from one hospital, in The Netherlands, and it is likely not correct to assume that the cohort is representative for all countries in the European Union.Sample collection was not constant in time as it was influenced by holidays and device downtime due to maintenance work.The collection started during the transitional period.We earlier published a dataset of patient samples collected and analyzed before the titanium dioxide ban, and reported a prevalence of 14%, although this was not a representative cohort [9].Due to the large average size of the particles, we suspect that the identified particles were A total of 302 samples from patients with swollen joints were analyzed with a H-iRPolM Raman Spectrometer for the presence of TiO 2 .All patients were recruited from the VieCuri Medical Centre in Venlo, The Netherlands.On average, samples were analyzed within 24 h after puncture (SD 1.5 day).The trend in prevalence was calculated with a 3-month MOVAG algorithm, which was then used to fit an exponential function ( f (x) = ae −bx+c ) using least squares regression.There was a significant difference in the prevalence of TiO 2 between the transitional period (February 2022-April 2022) and the period during the ban (p < 0.01).

Discussion
We here present a decrease in prevalence of TiO 2 nanoparticles in synovial fluid samples drawn from Dutch patients with swollen joints.This decrease overlaps in time with the new EU regulations banning the use of TiO 2 nanoparticles in food products.Although TiO 2 was never identified in patients with active gout, simultaneous presence of TiO 2 particles and calcium pyrophosphate crystals was identified in two patients.The question rises whether a synergistic effect can be attributed to CPP and TiO 2 , as both particles can induce arousal of the NLRP3 inflammasome [15,18].The combination of two pro-inflammatory particles in the same inflamed joint is known as mixed crystal disease [19], a condition mostly attributed to the simultaneous presence of MSU and CPP crystals.As this is a rare phenomenon, there is little known about the severity of the disease and optimal disease management.
This study is not without limitations.Raman spectroscopy is a qualitative method and relies on microscopic localization of particles.Optical microscopic platforms are limited in resolution to the diffraction limit and have difficulties distinguishing objects smaller than ~300 nm.Furthermore, we analyzed only a small volume (40 µL) of fluid, whilst synovial fluid punctures can retrieve up to 50 mL of material.The patients were all recruited from one hospital, in The Netherlands, and it is likely not correct to assume that the cohort is representative for all countries in the European Union.Sample collection was not constant in time as it was influenced by holidays and device downtime due to maintenance work.The collection started during the transitional period.We earlier published a dataset of patient samples collected and analyzed before the titanium dioxide ban, and reported a prevalence of 14%, although this was not a representative cohort [9].Due to the large average size of the particles, we suspect that the identified particles were most probably congregates of clustered nanoparticles.Individual particles are often smaller (<100 nm) and therefore can be missed in our analysis.Strong points of this study include the inclusion of blank samples, which increases the confidence that there is a small chance for false positives.Furthermore, a large number of patients were included and there was no selection of patient samples, because we analyzed all available samples from our outpatient clinic.TiO 2 is not only used for whitening of some food products but are widely used in other products like cosmetics, sunscreens, paints, and prosthetic implants.Therefore, it is expected that a ban of TiO 2 in only food products will not result in a zero occurrence of TiO 2 in humans.While scientific research has demonstrated the potential harmful effects of TiO 2 nanoparticles as a food additive, their use is still an active point of debate.With proper regulatory measures, the imposed threat of TiO 2 nanoparticles to public health can be mitigated.This research demonstrates how regulation can lead to a measurable effect in prevalence of potentially toxic particles.

Figure 1 .
Figure 1.Examples of spectra from TiO2 particles measured in patient samples.The upper spectrum is identified as anatase and was measured in a sample from a patient diagnosed with rheumatoid arthritis of the hand.The lower spectrum is identified as rutile and was measured in a sample from a patient with an undiagnosed swollen knee joint.

Figure 1 .
Figure 1.Examples of spectra from TiO 2 particles measured in patient samples.The upper spectrum is identified as anatase and was measured in a sample from a patient diagnosed with rheumatoid arthritis of the hand.The lower spectrum is identified as rutile and was measured in a sample from a patient with an undiagnosed swollen knee joint.

Figure 2 .
Figure 2. The prevalence of TiO2 in synovial fluid samples between 1 April 2022 and 15 June 2023.A total of 302 samples from patients with swollen joints were analyzed with a H-iRPolM Raman Spectrometer for the presence of TiO2.All patients were recruited from the VieCuri Medical Centre in Venlo, The Netherlands.On average, samples were analyzed within 24 h after puncture (SD 1.5 day).The trend in prevalence was calculated with a 3-month MOVAG algorithm, which was then used to fit an exponential function (  =  ) using least squares regression.There was a significant difference in the prevalence of TiO2 between the transitional period (February 2022-April 2022) and the period during the ban (p < 0.01).

Figure 2 .
Figure 2. The prevalence of TiO 2 in synovial fluid samples between 1 April 2022 and 15 June 2023.A total of 302 samples from patients with swollen joints were analyzed with a H-iRPolM Raman Spectrometer for the presence of TiO 2 .All patients were recruited from the VieCuri Medical Centre in Venlo, The Netherlands.On average, samples were analyzed within 24 h after puncture (SD 1.5 day).The trend in prevalence was calculated with a 3-month MOVAG algorithm, which was then used to fit an exponential function ( f (x) = ae −bx+c ) using least squares regression.There was a significant difference in the prevalence of TiO 2 between the transitional period (February 2022-April 2022) and the period during the ban (p < 0.01).

Table 1 .
Overview of patient characteristics in TiO 2 negative and the TiO 2 positive groups.
Significance was calculated with chi-square (for clinical diagnosis, sex, analyzed joint or bursa, presence of orthopedic implants, C-reactive protein levels) and a two tailed student t-test (Age).* First metatarsophalangeal joint, ** Other include hip, shoulder, elbow, finger.